Replica State - Error After Successful Blue/Green Switchover
We recently upgraded our RDS Postgres instances using blue/green deployment, from 16.3 to 18.3 (latest). Everything went smooth and was able to switchover successfully, we checked everything and all seems fine except we noticed that under Replication window it has error, see attached image.
Before this, we upgraded already 1 instance last month and there is nothing like this on that instance. I checked again anything and the DB looks healthy, no error on logs as well.
I cannot find any documentation related to this, not sure if this is expected behavior that we can just ignore since it make sense that after blue/green switchover the replication between new blue and old blue will break.
Is there someone experience something like this? or any idea if what I am thinking is correct and we can just safely ignore this for a while? TIA
- Newest
- Most votes
- Most comments
What you're observing is expected behavior and can be safely ignored. After a successful blue/green deployment switchover, replication between the blue and green environments stops. This is by design.
During the switchover process, the green environment becomes your new production environment, while the blue environment (your previous production) becomes the old environment. The DB instances in the previous blue environment are retained but are no longer actively replicating. Amazon RDS renames these instances and sets them to read-only mode.
The replication error you're seeing in the console is simply reflecting that the replication relationship that existed during the blue/green deployment process has been terminated, which is the intended outcome after switchover. Since you've confirmed that your database is healthy, there are no errors in the logs, and everything is functioning properly, this replication error status is just a residual indicator and doesn't represent an actual problem with your current production database.
The old blue environment instances are kept available in case you need to reference them or roll back, but they're no longer part of an active replication setup. You can eventually delete these old instances once you're confident you no longer need them for reference or rollback purposes.
Sources
Switching a blue/green deployment in Amazon RDS - Amazon Relational Database Service
Hi
We saw similar error before switchover after deployment blue/green for postgresql from 15 to 16. Same everything was good , out of 5 replication slot only 1 was inactive and was getting same message as Error in replication tab.
Error was related to max_logical_replication_workers and max_worker_processes 50 and 20 and rebooted green env which resolved the issue. This error was in green postgresql logs.
This resolved issue.
Thanks kumaran
Relevant content
- asked 2 years ago
- asked 2 years ago
- asked 2 years ago
- AWS OFFICIALUpdated 9 months ago
Thank you for the very well explained answer Florian, just want to ask one more thing related to this.
So I tried again yesterday on my account to test if I can replicate the error, I created RDS Postgres instance with 16.3 version (no data), upgraded it through blue/green deployment again using latest version (18.3), then after the switchover there wasn't anything on the replica state, no error and no value at all.
What do you think is the possible reason why on our previous upgrade it has error on replica state? Did we miss something or its just this can happen expectedly as you've said and nothing we should worry about as long as DB is healthy and no error on logs at all. We just want to make sure every steps are correct since we planning to do it on our Prod servers by next month.